www.elsevier.com/locate/cattod

Catalysis Today 123 (2007) 77–85

Synthesis of vinyl acetate on Pd-based catalysts

D. Kumar, M.S. Chen, D.W. Goodman *

Department of Chemistry, Texas A&M University, P.O. Box 30012, College Station, TX 77842-3012, United States

Available online 12 March 2007

Abstract

Vinyl acetate (VA) synthesis over Pd–Au catalysts is an important industrial reaction that has been studied extensively; however, there is no

consensus regarding the reaction mechanism, the active site, the key intermediates, and the role of Au. Recent results from our laboratories using a

combination of surface science and kinetic methods on technical and model catalytic systems have established that the VA synthesis reaction is

structure sensitive, including being dependent on the Pd–Au particle size. The role of Au is to isolate surface Pd atoms into Pd monomeric sites

thereby enhancing the VA formation rate and selectivity. This paper reviews the current understanding of this reaction on Pd, Pd–Au, and Pd–Sn

catalysts.

# 2007 Elsevier B.V. All rights reserved.

Keywords: Vinyl acetate; Pd; Pd–Au; Pd–Sn

1. Introduction

Vinyl acetate (VA) is an important chemical intermediate

utilized in paints, adhesives and surface coatings. The oldest

VA synthesis process is the gas-phase acetoxylation of

acetylene, based upon vapor phase reaction over a zinc acetate

catalyst, typically supported on carbon. The reaction yields are

generally very high ranging from 92 to 98% based on acetylene.

However, the expense and scarcity of acetylene made this

process unattractive, and in the early 1960s, several processes

evolved using ethylene as a feedstock, such as the homo-

geneous catalytic acetoxylation of ethylene over Pd and Cu

chloride catalysts [1–4]:

C2H4þ PdX2þHX ! ½C2H4PdX3�� þHþ;

½C2H4PdX3�� þAcOH ! ½C2H4PdX2ðOAcÞ�� þHX;

½C2H4PdX2ðOAcÞ�� ! ½C2H4PdXðOAcÞ�� þX�;

½C2H4PdXðOAcÞ� ! CH2CH�OAc þ Pd þ HX ðX ¼ ClÞ

Later, this process was replaced with gas phase acetoxylation of

ethylene over a Pd–Au bimetallic silica-supported catalyst

promoted with potassium acetate (KOAc) at a reaction tem-

* Corresponding author. Tel.: +1 979 845 0214; fax: +1 979 845 6822.

E-mail address: goodman@mail.chem.tamu.edu (D.W. Goodman).

0920-5861/$ – see front matter # 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.cattod.2007.01.050

perature of 423–463 K and a total pressure of 600–1000 kPa

[5–11]:

CH3COOH þ C2H4þ 0:5O2 ! CH3COOCHCH2þH2O

Since, the late 1960s when the gas phase reaction of ethylene

and acetic acid in the presence of oxygen to form VA over Pd-

based/SiO2 catalyst was first reported, there has been

considerable interest in this reaction from the industrial sector.

Recently academic researchers have shown an intense interest

in investigating the mechanism of the synthesis of VA [7,12–

16], although the basic aspects of the reaction were studied in

the early 1970s [3,5,9]. Two commercial processes, both

employing Pd-based catalysts, have been developed for the

production of VA. The homogenous liquid phase process

accounts for approximately 25% of the total VA production,

while the heterogeneous gas phase process accounts for the

remainder.

Although considerable work addressing the kinetic and

mechanistic parameters has been reported on various Pd-based

catalysts, there is no consensus regarding the reaction

mechanism and the nature of the active sites/intermediates in

VA synthesis. In this article, recent developments from our

laboratories with respect to VA synthesis are summarized with

an emphasis on our current understanding of the active site and

the reaction mechanism.

Fig.2. VAsynthesisonPd(5.0 wt.%)/SiO2,Pd(1.0 wt.%)/SiO2andPd(1.0 wt.%)–

Au(0.4 wt.%)/SiO2 catalysts. pO2¼ 7:6 Torr; pC2H4

¼ 57:0 Torr;pAcOH = 12.0 -

Torr; the remainder N2; flow rate: 30–60 ml/min; temperature = 413 K; catalyst

weight: 0.1–1.6 g [18].

D. Kumar et al. / Catalysis Today 123 (2007) 77–8578

2. Pd-only catalysts

In 1960, Moiseev et al. investigated the homogenous liquid-

phase production of VA using a [PdCl2(C2H4)]2 complex

catalyst [3]. The reaction was studied by vapor phase oxidation

of ethylene and acetic acid by Somanos et al. in 1971 on Pd

catalysts supported by SiO2 and Al2O3 [9]. In this study, various

parameters such as the effect of Pd dispersion, the partial

pressures of the reactants/water were evaluated. Within this

same time period, Nakamura and Yasui studied the mechanistic

aspects of VA synthesis on Pd-based catalysts [5], investigating

the promotional effect of alkali metals and the reaction orders

with respect to acetic acid and ethylene. Furthermore, the

heterogeneous process for VA production was shown to be

preferable due to limited corrosion and the absence of

chlorides.

Using Pd/SiO2 catalysts, approximately 80–90% selectivity

has been reported for VA synthesis [9,14]. In our laboratories,

Pd(5.0 wt.%) and Pd(1.0 wt.%)/SiO2 catalysts prepared by wet

impregnation methods [13] were shown to have an average

particle size after reduction of (4.2 � 0.2) and (2.5 � 0.1) nm,

respectively, using transmission electron microscopy (TEM).

For the same catalysts X-ray diffraction (XRD) indicated the

average particle size to be 3.8 and 2.8 nm, respectively. Under

reaction conditions TEM showed the average particle size to

increase to 5.5 and 3.5 nm for Pd(5.0 wt.%) and Pd(1.0 wt.%),

respectively. The catalysts maintained a steady activity after an

initial induction period within the first 100 min. Deactivation

was very slow and mainly attributed to sintering and to the

formation of surface carbide [13,17]. As shown in Fig. 1, the

formation of PdCx was dependent on the particle size of the

catalyst. The Pd(1.0 wt.%)/SiO2 catalyst formed less carbide on

the surface in comparison to the Pd(5.0 wt.%)/SiO2 catalyst.

The reaction rates and selectivities were higher on the 1.0 wt.%

Pd with smaller particle size than on the 5.0 wt.% Pd with a

larger average particle size, as reflected in Fig. 2 [18]. The

dependence of the reaction rate and selectivity on the particle

size indicates a degree of structure sensitivity. On the Pd(1 0 0)

the rate was lower confirming that larger particles are relatively

less active for VA formation [18]. The corresponding activation

Fig. 1. XRD data for Pd-only catalysts. (a and b) Pd(5.0 wt.%)/SiO2; (c and d)

Pd(1.0 wt.%)/SiO2; (a and c) freshly reduced; (b and d) after reaction [13].

energy (Ea) was higher (39.0 kJ/mol) for the catalyst with

smaller Pd particles than for the catalyst with larger Pd particles

(17.3 kJ/mol). Typically, on promoted 2.0 wt.% Pd/SiO2 and

5.0 wt.% Pd/Al2O3 catalysts, the Ea was near 30.0 kJ/mol [5,9].

The reaction orders with respect to ethylene and oxygen

were reported to be negative on 2.0 wt.% Pd/SiO2 promoted

with sodium acetate [9], and positive on 1.0 and 5.0 wt.% Pd/

SiO2 catalysts [13] and on 5.0 wt.% Pd/Al2O3 between 50 and

200 kPa [5]. A Langmuir–Hinshelwood mechanism was

suggested based on the observed positive reaction order with

respect to O2 and negative order with respect to C2H4. The

reaction orders with respect to ethylene and oxygen were

independent of the particle size. The reaction order with respect

to acetic acid has been a point of contention with reported

values ranging from 0 to 1 [9,13,19]. On Pd(5.0 wt.%)/SiO2 and

Pd(1.0 wt.%)/SiO2 catalysts, we found that the reaction order

with respect to acetic acid is almost zero [13].

The catalytic stability was strongly dependent on the particle

size with Pd(1.0 wt.%)/SiO2 showing less deactivation com-

pared to other catalysts after an initial induction period [18].

Overall, the kinetic parameters on Pd(1 0 0) and on supported

Pd catalysts suggest a similar mechanism for VA formation

[18].

3. Catalysis by Pd–Au

The reactivity of metal surfaces is a critical function of

composition and structure with alloys often showing unique

properties compared to the corresponding single component

metals [20–22]. For example, mixtures of Pd and Au are used as

catalysts for a number of applications [23–26] including

hydrogen fuel cells [27] and pollution control [28]. The

addition of Au to Pd significantly enhances the catalytic

activity, selectivity and stability; however, the details of this

promotional effect are not well understood.

D. Kumar et al. / Catalysis Today 123 (2007) 77–85 79

Au–Pd alloys are completely miscible, with the heats of

formation of Au–Pd alloys at 300 K being negative over the

entire composition range; maximum stability occurs at �60%

Au [24]. The observed large negative enthalpies are indicative

of an attractive interaction between the alloy constituents and a

tendency to order. The lattice parameters of Au–Pd alloys at

300 K vary from 0.389 nm for pure Pd to 0.408 nm for pure Au

with a small negative deviation from Vegard’s law [29]. Short-

range ordering in Au–Pd alloys has been reported and reviewed

by Allison et al. [24]. Early work on the properties of these Pd–

Au catalysts concluded that the 0.6 d-band vacancies of pure Pd

fill at an alloy composition of �60% Au, a results that

stimulated studies of the catalytic properties of Au–Pd mixtures

at this critical composition, e.g., ortho, para-hydrogen

conversion [30]. However, in surveying a number of reactions,

the consensus is that surface composition is more important for

adsorption and catalysis than is the bulk electronic structure

[12,24,31–34]. There is still considerable uncertainty in the

interpretation of the surface properties, since the surface

composition can differ markedly from the bulk, and the surface

atoms can be quite mobile under reaction conditions. In any

case, theoretical calculations have shown a correlation among

surface strain, adsorption energy, and activation barriers for

reaction [35].

3.1. Supported catalysts

Debate regarding the catalytic active site for VA synthesis

has focused on whether Pd is present as Pd0 or Pd2+. Moiseev

et al. have argued that the active site consists of Pd0, which

catalyze very selectively the vinyl acetate reaction [3]. More

convincingly, the active site was proposed to contain an acetate

species [36,10,5], Pd(OAc)2, formed from acetic acid adsorp-

tion [19]. The proposed reaction mechanism is [10]:

C2H4þ 2Pd , CH2�CH�Pd þ PdH;

O2þ 2Pd , 2Pd�O;

CH3COOH þ Pd , Pd�CH3COOH;

Pd�CH3COOH þ Pd�O , Pd�O�COCH3þ Pd�OH;

Pd�O�COCH3þCH2CH�Pd, Pd�CH2CHOCOCH3þ Pd;

Pd�CH2CHOCOCH3 , Pd þ CH2CHOCOCH3;

Pd�OH þ PdH , Pd�OH2 , Pd þ H2O

It is probable that a gas–solid mechanism is involved in this

reaction. However, there is strong evidence showing that a

liquid phase-like mechanism is also occurring under reaction

conditions. Under typical industrial operating conditions, the

absorption of acetic acid and water on the support can be

substantial [9]. An absorbed acetic acid liquid layer, with

approximately three monolayers in thickness at temperatures

comparable to industrial conditions on Pd/Cd/silica and Pd/Au/

silica catalysts, was evident in isotopic transient kinetics and

TPD studies [37]. The presence of Pd and alkali metal acetate

(KOAc in particular) enhances the adsorption of acetic acid.

KOAc, known to form a dimer species with acetic acid, is

essential for a liquid phase [38] in immobilizing the acetic acid,

which, in turn, increases the reaction rate for vinyl acetate and

suppresses direct ethylene combustion [6,9].

Based on Monte Carlo simulations, Neurock et al. [39,15]

have proposed that the critical ensemble required to couple

ethylene and acetate to VA consists of several Pd atoms. The

addition of Au improves the activity for vinyl acetate synthesis by

approximately a factor of 2 and the selectivity by approximately

5%. Comparative kinetic studies of VA synthesis were carried out

for Pd(1.0 wt.%)/SiO2 and Pd(1.0 wt.%)–Au(0.5 wt.%)/SiO2

catalysts, with the highly dispersed metal particles (Pd and Pd–

Au) characterized by XRD and TEM-EDS [14]. For Pd–Au

catalysts, the reaction orders with respect to C2H4 and oxygen are

both positive. High surface area SiO2 supported Pd and Pd–Au

VA synthesis catalysts show that addition of approximately

20 at.% of Au enhances the VA formation rate by a factor of 16 as

shown in Fig. 2. This change in the kinetic parameters is

attributed to a change in the electronic and geometric properties

of Pd upon alloying with Au. Confirming the observations of

Macleod et al. [7] a reduction in the Pd concentration leads to an

increase in the Au surface concentration. The formation of Pd

carbide (PdCx) during the synthesis of VA was also investigated

over a Pd–Au/SiO2 mixed-metal catalyst [17]. XRD and XPS

data showed no PdCx species at the surface of the Pd–Au catalyst,

i.e., alloying of Au with Pd is very effective in preventing PdCx

formation in Pd-based catalysts for VA synthesis.

An early study directly addressing the role of Au in

enhancing the VA reaction rate is that by Provine et al. [6].

While noting that there is indeed an increase in the production

of VA, it was reported that the stability of certain surface

species was altered with the addition of Au and that Au

suppresses the formation of CO via combustion of acetic acid.

The addition of Au also enhanced the production of VA by

facilitating product desorption. While corroborating this,

Lambert and coworkers further proposed that Au inhibits the

decomposition of VA [7]. The addition of Au also was

purported to enhance the formation of a monodentate surface

acetate species at 1735 cm�1. This species, proposed as a key

reaction intermediate, was observed under reaction conditions

on a working Pd catalyst by Augustine and Blitz [19].

The commercial VA synthesis catalyst, Pd–Au–K/SiO2,

contains approximately 1 wt.% of Au and Pd with a 4:1 Pd:Au

atomic ratio, and 2.5 wt.% of potassium acetate [7,14]. For a

commercial Pd–Au–K/SiO2 catalyst, progressive loss of

activity occurs during operation, restricting the lifetime.

Detailed XRD and HREM/EDX investigations [7] show that

fresh catalysts contain Pd in two forms: Pd–Au alloy particles

4–5 nm in diameter with a Pd:Au ratio of 2:3 and a very highly

dispersed Pd metal component corresponding to �85% of the

total Pd loading. As a consequence, the alloy particles are far

more Au-rich (60 at.% Au) than expected from the metal

loadings (�20% Au). After aging (deactivation), pronounced

sintering of the Pd–Au alloy particles and the appearance of Pd

acetate were observed, but without significant change in

composition of the alloy particles. It was found that high

reaction temperatures as well as high oxygen partial pressures

favor deactivation while both the ethylene concentration and

initial GHSV influence very little the deactivation rate [40].

Fig. 3. Surface vs. bulk compositions for Au–Pd polycrystalline alloys (A) (*) steady-state sputtered surfaces, (*) annealed at 973 K [43], and (B) for Au–Pd thin

films on Mo(1 1 0) [45].

Fig. 4. Surface concentrations of Pd and Au of 5 ML Pd/5 ML Au/Mo(1 1 0)

(*) and 5 ML Au/5 ML Pd/Mo(1 1 0) (&) as a function of annealing tem-

peratures [45].

D. Kumar et al. / Catalysis Today 123 (2007) 77–8580

3.2. Surface versus bulk composition

A number of studies have shown that the surface

composition of Au–Pd mixtures differs from the respective

bulk composition, with the surface being enriched in Au [41–

44]. Surface enrichment in Au is consistent with Au having a

much lower surface free energy compared with Pd. AES studies

of Au–Pd alloy surfaces containing various bulk compositions

show that Au is significantly enriched in an annealed surface

[42]. These results have been confirmed using low-energy ion

scattering spectroscopy (LEIS), a technique particularly

sensitive to the top-most surface layer, as shown in

Fig. 3(A) [43]. The effect of annealing temperatures on the

surface composition of Pd and Au (5 ML Pd/5 ML Au/

Mo(1 1 0) and 5 ML Au/5 ML Pd/Mo(1 1 0)) was investigated

using LEIS as shown in Fig. 4 [45]. After annealing to higher

temperature inter-diffusion of Pd–Au is evident. For example,

at 800 K the surface is significantly enhanced in Au relative to

the bulk. Between 700 and 1000 K, a stable surface alloy

composition of Pd0.2Au0.8 is formed for a 50–50 bulk mixture

of Pd and Au, independent of the deposition sequence of the Pd

and Au. LEIS data were obtained at various Pd and Au bulk

ratios to produce a Pd–Au surface phase diagram. As shown in

Fig. 3(B) the LEIS data for Au–Pd thin films supported on a

refractory single crystal Mo(1 1 0) surface are in excellent

agreement with the corresponding data measured for bulk

alloys (Fig. 3(A)) [42–44].

There are only a few examples where Au–Pd alloy single

crystal surfaces have been examined using a surface-layer

sensitive technique [46,47]. LEIS studies of Au3Pd(1 1 3) [47]

and Au3Pd(1 0 0) [46] surfaces show these surfaces to consist

essentially of a Au layer; Pd is found only within the second

atomic layer. Preferential surface segregation of one compo-

nent is common in metal alloy surfaces [48,49] and, in

particular, all Au-containing alloys show preferential surface

segregation of Au [49–51].

Based on the surface stabilities of Au–Pd alloys, epitaxial

growth of a Pd monolayer on Au(1 1 1) should be thermo-

dynamically unstable even at rather moderate temperatures.

Koel and coworkers [52] carried out detailed investigations of

Pd overlayers on Au(1 1 1) using AES and LEIS. Following the

deposition of Pd at 150 and 300 K, results consistent with a van

der Merwe growth mode, i.e., layer-by-layer. At 500 K,

however, the Pd AES intensity is markedly smaller while the

Au AES value is larger than the corresponding values at 150

and 300 K. These observations qualitatively indicate alloying

or inter-diffusion of a Pd monolayer at 500 K. Lambert and

coworkers [53] have addressed Pd submonolayers on

Au(1 1 1)-(22 �ffiffiffi

3p

), where Pd islands initially nucleate and

grow at the elbows near the surface edge dislocations of the Au

herringbone reconstruction. A morphological evolution of

these islands with increasing Pd coverage was observed and

provides an excellent explanation for their catalytic behavior.

Maroun et al. [54] using high resolution STM images

obtained from Pd–Au alloy surfaces prepared by electro-

chemically co-deposition of Pd and Au onto Au(1 1 1), clearly

resolved two types of atoms with different apparent heights and

Fig. 5. In situ atomic resolution STM images of PdAu alloys electrodeposited on Au(1 1 1) for (A) Pd07Au93 and (B) Pd15Au85. Pd atoms appear larger and,

depending on tunneling conditions, brighter or darker than Au atoms. (C) Surface coverages of Pd monomers, dimers, and trimers, as obtained from atomically

resolved STM images [54].

D. Kumar et al. / Catalysis Today 123 (2007) 77–85 81

shapes, arranged in a hexagonal lattice with a lattice spacing

equal to that of Au(1 1 1) (see Fig. 5). Varga and Hetzendorf

[44] also resolved two types of atoms with apparent chemical

contrast on Au3Pd(1 0 0) surface (see Fig. 6). The formation of

surface isolated Pd sites together with some dimers and trimers

were evident. These STM images with atomic contrast clearly

show that a single, uniform Au–Pd alloy phase with a

disordered metal-atom arrangement can be formed without

phase separation into Pd and Au domains.

Recent studies in our laboratories using infrared reflection

absorption spectroscopy (IRAS) with CO as a probe molecule

have clearly demonstrated the formation of isolated Pd surface

sites surrounded by Au on Au–Pd thin films supported on a

Mo(1 1 0) surface [45] and for Pd on Au(1 1 1) [12]. As shown

in Fig. 7, at room temperature after deposition of Pd on

Au(1 0 0) and Au(1 1 1), CO vibrational features correspond-

ing to adsorption on the twofold bridge and threefold hollow

sites are evidenced [45]. These results are consistent with

previous reports confirming the formation of Pd overlayers on

Au(1 1 1) and Au(1 0 0). After annealing these samples to

600 K or higher, the CO features corresponding to bridging and

threefold hollow sites disappear. This is also accompanied by

an increase in the intensity of the feature corresponding to CO

adsorption on atop sites of Pd. These results demonstrate that

much of the deposited Pd atoms diffuse into the bulk of the

catalysts forming a Pd–Au alloy with a limited number of Pd

atoms isolated as Pd monomers, i.e., Au4Pd on Au(1 0 0) and

Au6Pd on Au(1 1 1), after annealing. Thus, the formation of

isolated Pd atoms surrounded by Au atoms was observed.

Furthermore, the concentration of isolated Pd monomers

depends on the Pd–Au bulk concentration and the annealing

temperature. However, it should be noted that in the reaction

conditions Pd may be pulled to the surface due to the adsorption

of acetic acid [55].

3.3. Model catalysts

From recent single crystal studies in our laboratory, details

regarding the Pd–Au ensemble required for VA synthesis have

been revealed [12]. In these studies Pd was vapor deposited

onto two gold substrates, Au(1 0 0) and Au(1 1 1), followed by

annealing for 10 min at 550 K. Infrared reflection absorption

Fig. 6. Atomic resolution STM images of Au3Pd(1 0 0) surface (left). Pd atoms appear larger and brighter as indicated by circles or ellipses. The surface Pd coverage

is 14%, in which the relative amounts of monomer, monomer pair, dimer and those ensembles of more than two contiguous Pd atoms are shown in the histogram

(right) [44].

D. Kumar et al. / Catalysis Today 123 (2007) 77–8582

spectroscopy (IRAS) and CO adsorption on Pd/Au(1 0 0) and

Pd/Au(1 1 1) surfaces confirmed that Pd atoms form monomers

within a Au-rich surface. These Pd/Au(1 0 0) and Pd/Au(1 1 1)

catalysts were used to investigate the rate of formation of VA.

The reaction rate increased to a maximum at a coverage of

0.1 ML of Pd on Au(1 0 0) and decreased with a further

increase in the Pd coverage (see Fig. 8) up to 1.0 ML at which

coverage the reaction rate is essentially the same as that for

Pd(1 0 0). For the optimum surface of 0.1 ML of Pd, the density

of Pd monomers is the highest, consistent with Pd monomers

being the active sites for the formation of VA on Pd/Au

surfaces.

Fig. 7. IRAS spectra after CO adsorption at 90 K on 4 ML Pd/Au(1 0 0) and

4 ML Pd/Au(1 1 1). Pd was deposited on Au(1 0 0) and Au(1 1 1) at 90 K and

subsequently annealed to 300 and 600 K for 10 min each [12].

The VA synthesis rate on a Au-only surface is insignificant

compared to the Pd-only surface, therefore it is then

straightforward to measure the VA reaction rate for the Pd–

Au alloy catalysts with respect to the atomic fraction of Pd on

the surface, as shown in Fig. 9. The formation rate per Pd on

Au(1 0 0) surface varies with the amount of Pd on the surface

with the maximum in the activity occurring at a coverage of

0.07 ML. While a decrease is apparent in the reaction rate with

an increase or decrease in the coverage of Pd on Au(1 0 0), a

quite different behavior is found following deposition of Pd

onto a Au(1 1 1) surface. A continuous increase in the reaction

rate is observed with a decrease in the surface coverage of Pd on

Au(1 1 1). These data for VA formation rate on Pd on Au(1 0 0)

Fig. 8. VA formation rate as a function of the Pd coverage on Au(1 0 0) and

Au(1 1 1) surfaces. TOFs are calculated with respect to (1 � 1) surface unit. Pd

was deposited on the respective surface at 300 K and annealed to 600 K for

10 min. C2H4 = 8 Torr, O2 = 2 Torr, CH3COOH = 4 Torr, reaction tempera-

ture = 453 K, reaction time = 3 h [12].

Fig. 9. Vinyl acetate (VA) formation rates as a function of Pd coverage. The

TOFs are computed with respect to the Pd atom concentration. Pd was deposited

on the respective surface at 300 K and annealed to 600 K for 10 min.

C2H4 = 8 Torr, O2 = 2 Torr, CH3COOH = 4 Torr, reaction tempera-

ture = 453 K, reaction time = 3 h. The error bars are based on background rate

data. The inserts show Pd monomers and monomer pairs on the Au(1 0 0) and

Au(1 1 1) [12].

Fig. 10. Schematic for VA synthesis from acetic acid and ethylene. The

optimized distance between the two active centers for the coupling of surface

ethylene and acetate species to form VA is 3.3 A. With lateral displacement,

coupling of an ethylene and acetate species on a Pd monomer pair is possible on

the Au(1 0 0), but unlikely on the Au(1 1 1) [12].

D. Kumar et al. / Catalysis Today 123 (2007) 77–85 83

and Au(1 1 1) surfaces are strong evidence for isolated Pd

atoms being the active sites for the reaction. The apparent

difference in the per Pd reaction rates for Au(1 0 0) and

Au(1 1 1) surfaces also indicates that a pair of non-contiguous

Pd monomer sites, rather than a single isolated site, is required

for VA formation.

Two different reaction mechanisms have been proposed for

VA synthesis. The adsorption and subsequent activation of

ethylene to form vinyl species which then couples with the

coadsorbed acetate species to form VA is one proposed

mechanism [5]. In a second mechanism, adsorbed ethylene

reacts with an adsorbed acetate nucleophile to form as

intermediate species, ethyl-acetate. This species then under-

goes b-CH elimination to form VA [9,15]. Both mechanisms

invoke coupling of acetate and ethylene species as the rate-

limiting step [5,9,15]. Hence a correlated pair of Pd sites is

required for the formation of VA. Considering the bond lengths

of adsorbed ethylene and acetate species, the optimized

distance between two active sites is 3.3 A. Au(1 0 0) defines the

distance between a pair of Pd monomers to be 4.08 A while

Au(1 1 1) defines this distance to be 4.99 A, a prohibitively

long distance for coupling of these two reactive intermediates.

The reaction rate therefore on Pd/Au(1 1 1) is much lower than

that on Pd/Au(1 0 0) as evidenced in Fig. 9. The bonding and

relative distances involved between reacting species is shown

schematically in Fig. 10.

The pair of isolated Pd sites, while aiding in the formation of

VA by providing the optimum required spacing for coupling of

the surface acetate and ethylene species, was also proposed to

suppress the formation of reaction by-products, such as CO and

CO2, thus improving the overall selectivity. CO is a by-product

or reaction intermediate in VA synthesis [15]. Since, CO binds

more tightly to sites containing contiguous Pd atoms compared

to an isolated Pd site (Fig. 11(A)) [45], CO is likely a more

effective poison for larger Pd ensembles. Hence, addition of Au

to Pd was found to significantly suppress carbon deposition via

reactant and product decomposition which, in turn, leads to a

reduction in dehydrogenation rates that favors selective

acetoxylation [17,56–59].

In addition to CO, CO2 is also produced as a by-product in

VA synthesis due to the combustion of various reactants and the

VA product [37,59]. In general, there has been a long-standing

debate as to the contributions of the individual reactants, i.e.,

ethylene and acetic acid, toward the formation of CO2. While

some researchers have concluded that CO2 is mainly produced

from ethylene [5,9], others have argued that acetic acid also

contributes to the formation of CO2 [37]. On Pd–Cd/SiO2 and

Pd–Au/SiO2 promoted with K+, Crathorne et al. [37] have

reported that acetic acid and ethylene contribute equally to the

formation of CO2. Using various types of catalysts, such as

unsupported Pd, Pd supported on Al2O3, with the latter

promoted with K, Nakamura and Yasui asserted that CO2 is

primarily produced from ethylene [5]. Also, Somanos et al.

reached a similar conclusion using Pd supported on Al2O3 and

SiO2 catalysts in the presence and absence of acetic acid [9].

Ethylene oxidation was investigated in the presence and

absence of acetic acid and VA on Pd(5.0 wt.%)/SiO2 catalyst at

conditions close to those used for VA synthesis [59]. In the

presence and absence of acetic acid (2.0 kPa) there was no

obvious change in the reaction kinetic parameters. It was

concluded that within the temperature range of 413–453 K and

at pressures ranging between 1.0 and 10.0 kPa of O2 and 5.0–

15.0 kPa of C2H4, CO2 is primarily derived from ethylene

combustion.

The formation of isolated Pd monomer sites influences the

adsorption properties of ethylene as well. Indeed, TPD has

shown that ethylene and acetic acid bond significantly less

Fig. 11. (A) TPD of CO on: (a) 10 ML Pd/Mo(1 1 0) and (b) 5 ML Pd/5 ML

Au/Mo(1 1 0) annealed to 800 K for 20 min [45,59]. (B) TPD of C2D4 from

supported Pd and Pd–Au bimetallic clusters [59].

Fig. 12. VA formation rate normalized to per Pd site basis on Sn/4.0 ML Pd/

Rh(1 0 0) annealed to 600 K for 60 s. Reaction at 450 K; pO2¼ 2:0 Torr;

pC2H4¼ 9:0 Torr; pAcOH = 4.0 Torr; time = 3 h [63].

D. Kumar et al. / Catalysis Today 123 (2007) 77–8584

strongly to a Pd monomer compared to a site containing

contiguous Pd atoms [60] (Fig. 11(B)). On continuous Pd sites a

di-s bonded ethylene species and an ethylidyne species were

observed [61]. The decomposition of these species leads to the

formation of carbon, whereas the addition of Au inhibits the

dissociation of ethylene, consequently forming less carbon and

extending the activity of the catalyst.

4. VA synthesis using Pd–Sn surfaces

The epitaxial growth of metals on refractive metal surfaces

has been used to study the properties of thin metal films. For

example, Pd and Sn grow layer-by-layer on several metal

substrates [62,63]. Recently Sn was vapor deposited onto a

4.0 ML thick film of Pd and annealed to 600 K for 1.0 min [64].

For VA reaction times of 3.0 h, a maximum rate expressed per

Pd was found after deposition of approximately 0.5 ML of Sn,

which is significantly higher relative to a Pd-only surface.

Further increase in the Sn coverage leads to a reduction in the

VA formation rate, as shown in Fig. 12. A Sn coverage of

0.5 ML corresponds to a c(2 � 2) Pd–Sn structure whose

formation has been observed for Sn on Pd(1 0 0) using low

energy electron diffraction (LEED) [65]. Similar results have

been reported by the Koel and Lambert groups on Pd(1 1 1),

Pd(1 1 0), Pt(1 1 1) and Pt(1 0 0) surfaces [61,62,66–68]. The

total density of Pd monomer pairs is maximized at this coverage

and the distance between each isolated Pd monomer is 3.89 A.

In fact, subsequent to the formation of the c(2 � 2) structure,

the distance between any two contiguous Pd atoms is identical.

Since it has been proposed that a pair of Pd monomers suitably

spaced act as the active sites [12], the VA formation rate should

be maximized for a c(2 � 2) Sn–Pd structure, as is observed.

With further increase in Sn coverage the VA formation rate

decreases slightly likely due to the decrease in the Pd monomer

density and the formation of Sn islands on the surface of the

catalyst. Excess Sn diffusing into the subsurface region of Pd

and a Sn overlayer could influence the adsorption and reaction

of neighboring Pd atoms and lead to a decrease in the overall

activity.

5. Conclusions

A combination of surface science and kinetic methods have

been used to study the role of Au in Pd–Au alloy catalysts for VA

synthesis. The reaction has been shown to be structure sensitive

including a strong dependence on metal particle size.

Furthermore, addition of Au to a Pd catalyst significantly

enhances the VA formation rate while reducing the amount of

carbon formed on the surface. LEIS was used to investigate the

surfaceversus bulk composition for Pd–Au surfaces. Using IRAS

and CO as a probe, Au has been shown to isolate surface Pd into

monomeric sites which have been proposed to be the active site

for the formation of VA. A pair of such isolated Pd monomers is

required for the adsorption of acetic acid and ethylene, with the

distance between the constituent monomers being critical.

Hence, while separating Pd atoms on the surface into monomer

sites, Au also reduces the formation of undesirable surface

D. Kumar et al. / Catalysis Today 123 (2007) 77–85 85

species and products. Utilization of Pd–Sn surfaces for the

formation of VA is further confirmation that isolated Pd

monomeric sites are indeed the active site for this reaction.

Acknowledgements

We greatly acknowledge the support of this work by the

Department of Energy, Office of Basic Energy Sciences,

Division of Chemical Sciences, and the Robert A. Welch

Foundation.

References

[1] G.S. Grover, R.V. Chaudhari, Chem. Eng. J. 32 (1986) 93.

[2] D.D. Kragten, R.A. van Santen, M.K. Crawford, W.D. Provine, J.J. Lerou,

Inorg. Chem. 38 (1999) 331.

[3] I.I. Moiseev, M.N. Vargaftic, Y.K. Syrkin, Dokl. Akad. Nauk Az. SSR 133

(1960) 377.

[4] C.R. Reilly, J.J. Lerou, Catal. Today 41 (1998) 433.

[5] S. Nakamura, T. Yasui, J. Catal. 17 (1970) 366.

[6] W.D. Provine, P. Mills, J.J. Lerou, Stud. Surf. Sci. Catal. 101 (1996) 191.

[7] N. Macleod, J.M. Keel, R.M. Lambert, Appl. Catal. A 261 (2004) 37.

[8] W. Schwerdt, Chem. Ind. 45 (1968) 1559.

[9] B. Samanos, P. Boutry, R. Montarna, J. Catal. 23 (1971) 19.

[10] S. Nakamura, T. Yasui, J. Catal. 23 (1971) 315.

[11] M.L. Luyben, B.D. Tyreus, Comput. Chem. Eng. 22 (1998) 867.

[12] M.S. Chen, D. Kumar, C.W. Yi, D.W. Goodman, Science 310 (2005) 291.

[13] Y.F. Han, D. Kumar, D.W. Goodman, J. Catal. 230 (2005) 353.

[14] Y.F. Han, J.H. Wang, D. Kumar, Z. Yan, D.W. Goodman, J. Catal. 232

(2005) 467.

[15] M. Neurock, J. Catal. 216 (2003) 73.

[16] D. Stacchiola, F. Calaza, F. Burkholder, W.T. Tysoe, J. Am. Chem. Soc.

126 (2004) 15384.

[17] Y.F. Han, D. Kumar, C. Sivadinarayana, A. Clearfield, D.W. Goodman,

Catal. Lett. 94 (2004) 131.

[18] D. Kumar, Y.-F. Han, M.S. Chen, D.W. Goodman, Catal. Lett. 106 (2006) 1.

[19] S.M. Augustine, J.P. Blitz, J. Catal. 142 (1993) 312.

[20] J.H. Sinfelt, Bimetallic Catalysts: Discoveries, Concepts and

Applications, Wiley, New York, 1983.

[21] R.L. Moss, L. Whally, Advances in Catalysis, Academic Press, New York,

1972.

[22] V. Ponec, Adv. Catal. 32 (1983) 149.

[23] R. Abel, P. Collins, K. Eichler, I. Nicolau, D. Peters, in: G. Ertl, H.

Knozinger, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis,

vol. 5, Wiley/VCH, Weinheim, 1997, p. 2298.

[24] E.G. Allison, G.C. Bond, Catal. Rev. 7 (1972) 233.

[25] P.N. Rylander, Catalytic Hydrogenation in Organic Synthesis, Academic

Press, London, 1979.

[26] A.M. Venezia, V. La Parola, B. Pawelec, J.L.G. Fierro, Appl. Catal. A 264

(2004) 43.

[27] D.L. Trimm, Z.I. Onsan, Catal. Rev. 43 (2001) 31.

[28] M. Bonarowska, A. Malinowski, W. Juszczyk, Z. Karpinski, Appl. Catal.

B 30 (2001) 187.

[29] A. Maeland, T.B. Flanagan, Can. J. Phys. 42 (1964) 2364.

[30] A. Couper, D.D. Eley, Discuss. Faraday Soc. 8 (1950) 172.

[31] C.J. Baddeley, R.M. Ormerod, A.W. Stephenson, R.M. Lambert, J. Phys.

Chem. 99 (1995) 5146.

[32] D.D. Eley, P. Luetic, Trans., Faraday Soc. 53 (1957) 1483.

[33] G. Reinacker, S. Engels, Z. Anorg. Allg. Chem. 336 (1965) 259.

[34] W.M.H. Sachtler, P. van Der Planck, Surf. Sci. 18 (1969) 62.

[35] M. Mavrikakis, B. Hammer, J.K. Norskov, Phys. Rev. Lett. 81 (1998)

2819.

[36] H. Debellefontaine, J. Besombes-Vailhe, J. Chim. Phys. 75 (1978) 801.

[37] E.A. Crathorne, D. MacGowan, S.R. Morris, A.P. Rawlinson, J. Catal. 149

(1994) 254.

[38] Handbook of Chemistry and Physics, 48th ed., Chemical Rubber, 1967.

[39] M. Neurock, W.D. Provine, D.A. Dixon, G.W. Coulston, J.J. Lerou, R.A.

Vansanten, Chem. Eng. Sci. 51 (1996) 1691.

[40] Q. Smejkal, D. Linke, U. Bentrup, M.-M. Pohl, H. Berndt, M. Baerns, A.

Bruckner, Appl. Catal. A 268 (2004) 67.

[41] S.M. Foiles, J. Vac. Sci. Technol. A 5 (1987) 889.

[42] A. Jablonski, S.H. Overbury, G.A. Somarjai, Surf. Sci. 65 (1977) 578.

[43] D.G. Swartzfager, S.B. Ziemecki, M.J. Kelly, J. Vac. Sci. Technol. 19

(1981) 185.

[44] P. Varga, G. Hetzendorf, Surf. Sci. 162 (1985) 544.

[45] C.W. Yi, K. Luo, T. Wei, D.W. Goodman, J. Phys. Chem. B 109 (2005)

18535.

[46] M. Aschoff, S. Speller, J. Kuntze, W. Heiland, E. Platzgummer, M.

Schmid, P. Varga, B. Baretzky, Surf. Sci. 415 (1998) L1051.

[47] M. Aschoff, G. Piaszenski, S. Speller, W. Heiland, Surf. Sci. 402–404

(1998) 770.

[48] H. Niehus, W. Heiland, E. Taglauer, Surf. Sci. Rep. 17 (1993) 213.

[49] H. Niehus, Phys. Status Solidi B 192 (1995) 357.

[50] S. Schomann, E. Taglauer, Surf. Rev. Lett. 3 (1996) 1823.

[51] L. Houssiau, P. Bertrand, Nucl. Instrum. Methods B 125 (1997) 328.

[52] B.E. Koel, A. Sellidj, M.T. Paffett, Phys. Rev. B. 46 (1992) 7846.

[53] A.W. Stephenson, C.J. Baddeley, M.S. Tikhov, R.M. Lambert, Surf. Sci.

398 (1998) 172.

[54] F. Maroun, F. Ozanam, O.M. Magnussen, R.J. Bhem, Science 293 (2001)

1811.

[55] T.G. Owens, T.E. Jones, T.C.Q. Noakes, P. Bailey, C.J. Baddeley, J. Phys.

Chem. B 110 (42) (2006) 21152.

[56] M. Bowker, C. Morgan, J. Couves, Surf. Sci. 555 (2004) 145.

[57] M. Bowker, C. Morgan, Catal. Lett. 98 (2004) 67.

[58] M. Bowker, C. Morgan, N. Perkins, R. Holroyd, E. Fourre, F. Grillo, A.

MacDowall, J. Phys. Chem. B 109 (2005) 2377.

[59] Y.F. Han, D. Kumar, C. Sivadinarayana, D.W. Goodman, J. Catal. 224

(2004) 60.

[60] K. Luo, T. Wei, C.W. Yi, S. Axnanda, D.W. Goodman, J. Phys. Chem. B

109 (2005) 23517.

[61] M.S. Chen, K. Luo, Z. Wei, Z. Yan, D. Kumar, C.-W. Yi, D.W. Goodman,

Catal. Today 117 (2006) 37.

[62] N. Tsud, T. Skala, F. Sutata, K. Veltruska, V. Dudr, S. Fabik, L. Sedlacek,

V. Chab, K.C. Prince, V. Matolin, Surf. Sci. 595 (2005) 138.

[63] A.F. Lee, C.J. Baddeley, M.S. Tikhov, R.M. Lambert, Surf. Sci. 373 (1997)

195.

[64] D. Kumar, M.S. Chen, D.W. Goodman, unpublished data.

[65] T. Wei, K. Luo, S. Axnanda, D.W. Goodman, in press.

[66] J.M. Peck, D.I. Mahon, B.E. Koel, Surf. Sci. 410 (1998) 200.

[67] M. Batzill, D.E. Beck, B.E. Koel, Surf. Sci. 466 (2000) L821.

[68] A.F. Lee, C.J. Baddeley, C. Hardacre, G.D. Moggridge, R.M.

Ormerod, R.M. Lambert, J.P. Candy, J.-M. Basset, J. Phys. Chem. B

101 (1997) 2979.